Edge Graph Reconstructed Weight Model for Pose Graph Optimization Algorithm

In simultaneous localization and mapping (SLAM) systems based on graph optimization, the presence of erroneous closed-loop edges can interfere with the convergence of the graph optimizer, leading to a decrease in optimization speed and thus reducing the accuracy and robustness of the SLAM system. Therefore, we propose a pose graph optimization algorithm based on edge graph reconstructed weight model for erroneous closed-loop edges (EGR-PGO), which effectively improves the robustness of PGO algorithm. The algorithm introduces an edge graph transformation model and uses PageRank algorithm to dynamically adjust the parameters of the weight function, thereby optimizing the weights of closed-loop edges. In each iteration process, the algorithm will remove the erroneous closed-loop edges again based on the change in residuals and the length of the closed-loop edges to reduce the interference of erroneous closed-loop edges on the optimization process. Finally, we conduct Monte Carlo experiments on the PGO dataset, and the experimental results verify the speed and robustness of the EGR-PGO algorithm, as well as its effectiveness in the presence of error-loop-closure edges.